People are increasingly interacting with computers and other electronic devices in new and interesting ways. One such interaction approach involves making a specified motion with respect to a device, which can be detected in a field of view of a camera on the device. Since cameras have limited fields of view, it can be desirable to utilize multiple cameras to attempt to increase the effective field of view to be able to capture a larger range of motion. Unfortunately, continually capturing and analyzing images using multiple cameras can come with significant processing and power requirements, which can be particularly undesirable for portable computing devices with limited resources.